Shahul Shereef
Founder at Kaggle
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolShahul Shereef is a San Francisco–based data science leader and founder with eight years of experience building end-to-end ML and NLP systems in startups and production. As co-founder of ragas (YC W24), he’s driving an open-source standard for evaluating LLM applications while contributing to OpenAssistant and other community projects. A Kaggle Grandmaster ranked in the top 20 among 100,000+ users, he blends competition-grade modeling with practical MLOps—implementing dataset conversions for instruction-following dialogue and audio augmentations (RandomCrop, Padding, SpliceOut) for deep learning. His background includes credit-underwriting NLP, TensorFlow model deployment on Google Cloud, and adding evaluation metrics like BertScore and EditScore to open-source toolkits. Curious and product-minded, he publishes work and code publicly (shahules786.github.io) and focuses on tooling that makes ML systems more testable and reproducible.
8 years of coding experience
3 years of employment as a software developer
CGPA 8.01, CGPA 8.01 at Govt.model engineering college
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (19)
pytorch10
sentence-transformers10
python10
testing10
machine-learning10
datasets10
jupyter-notebook10
nlp10
data-augmentation10
natural-language-processing9
deep-learning9
language-model8
audio8
machinelearning8
language-models8
Programming languages (10)
Github contributions (5)
explodinggradients/ragas
May 2023 - Apr 2025
Supercharge Your LLM Application Evaluations 🚀
Role in this project:
Back-end Developer
Back-end DeveloperContributions:255 reviews, 582 PRs, 319 pushes in 1 year 10 months
Contributions summary:Shahul implemented a BertScore metric and added SBERT score calculation and relative imports within the `belar/metrics/similarity.py` file. Moreover, the user added EditScore metric with distance and ratio measures, and also a Bleu score. They also added Textual Entailment Score, fixed device checks, re-formatted imports and added the Q-square metric.
evaluationllmllmops
asteroid-team/torch-audiomentations
Mar 2022 - May 2022
Fast audio data augmentation in PyTorch. Inspired by audiomentations. Useful for deep learning.
Role in this project:ML Engineer
ML EngineerContributions:5 reviews, 52 commits, 4 PRs in 1 month
Contributions summary:Shahul primarily contributed to implementing and testing audio data augmentation techniques using PyTorch within the `torch-audiomentations` repository. They developed a `RandomCrop` augmentation, including initial implementation, base class initialization, type conversions, and testing. Furthermore, the user added a `Padding` augmentation and contributed to a `SpliceOut` augmentation. The commits focus on enhancing the library's audio processing capabilities, particularly for deep learning applications.
pythondifferentiable-data-augmentationaudio-datadspaudio-effects
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.